A building site electromechanical energy-saving linkage regulation and control system and method

Abstract

The application relates to the technical field of automatic control, and discloses a building site electromechanical energy-saving linkage regulation and control system and method, which comprises a data acquisition unit, a global mutual covariance matrix analysis unit, a control routing topology partition reconstruction unit, a logic rack bounce amount compensation unit and a control instruction output boundary unit; by acquiring return deviation change rate data flow and corresponding communication time stamps, the mutual covariance matrix is calibrated and constructed to identify coupling affinity, according to which the internal routing topology partition is reconstructed, and the bounce amount is determined according to the output amount difference at the moment of routing recombination and a limiting signal is sent; the application dynamically recombines the control routing topology in the virtual information space, eliminates the nonlinear coupling interference caused by the linkage of the controlled object, in-situ hedge phase coherence, smooths the topology switching transient state, suppresses the control saturation and commutation wear of the driving mechanism, and improves the anti-interference steady-state control capability of the system.

Description

Technical Field

[0001] This invention relates to a building site electromechanical energy-saving linkage control system and method, belonging to the field of automatic control technology. Background Technology

[0002] Currently, the use of independent closed-loop control loops to maintain the steady-state operation of a single power unit is a general control architecture. The system control center collects transient measurement signals from specific channels, calculates the control gain based on the control algorithm inside the control loop, and drives the actuator to converge the control variables to the preset target, thus realizing distributed control. The control command flow drives the actuator to regulate the fluid or pressure state, the sensor captures the physical state of the pipeline network and generates corresponding feedback voltage, the data acquisition interface receives the feedback voltage for the processing unit to calculate the deviation, and the control center corrects the subsequent control gain based on the deviation, so that the control variables of different control loops remain linear and independent under ideal conditions.

[0003] To address the aforementioned gain interference, conventional improvement schemes employ external high-precision sensors to acquire cross-interference quantities or increase the data sampling frequency to capture transient interference. However, adding external sensors leads to high hardware costs and increased system maintenance complexity. Simply increasing the sampling frequency can easily transform high-frequency step impacts into large-scale signal noise, causing the controller to generate high-frequency overmodulation, resulting in abrupt distribution of control commands. This accelerates commutation fatigue and zero-point temperature drift in the actuator. In the aforementioned electromechanical equipment group control scheme, not only are there physical limitations at the hardware level, such as unreconcilable rigid deformation and mechanical wear, but the software level also has significant deficiencies in dynamic multi-source parameter decoupling and command timing control. For example, Chinese invention patent CN122239504A discloses an energy-saving method and system for electromechanical equipment group control based on edge computing. The system attempts to use correlation coefficient and partial correlation coefficient matrices to calculate the influence weight of environmental parameters on air conditioning and lighting equipment, and then calculate the weighted deviation and generate adjustment commands. However, the existing technology implicitly relies on an idealized environmental state with synchronous sensor sampling and no communication delay. When facing the complex and non-ideal industrial high-noise environment of the building site, the objectively existing communication delay difference components and timing misalignment between different links can easily cause phase drift in the static linear fitting in the partial correlation analysis. This not only causes the calculated coupling affinity weight to deviate from the actual working condition, but also makes it impossible to eliminate the nonlinear coupling interference caused by the linkage of controlled objects when multiple devices are adjusted in parallel due to the lack of dynamic route reconstruction and output differential limiting mechanism in the virtual space. This causes the logic rack to generate command bounce at the moment of topology switching, resulting in frequent control saturation and chatter damage to the controlled drive mechanism.

[0004] Therefore, the technical problem to be solved by this invention is how to accurately identify the transient coupling strength between multiple adjustment loops and realize the internal offsetting of nonlinear interference between loops by dynamically reorganizing the control routing topology, thereby eliminating the sudden change in control commands during topology switching. Summary of the Invention

[0005] To address the problems in the background art, the technical solution of the present invention is as follows: A building site electromechanical energy-saving linkage control system, comprising: The data acquisition unit is used to acquire the feedback error change rate data stream and the corresponding communication timestamp of each control loop; The global covariance matrix analysis unit, connected to the data acquisition unit, is used to calibrate the phase of the feedback error change rate data stream based on the delay difference component of the communication timestamp, and to construct the covariance matrix of the feedback error change rate to identify the coupling affinity between control loops. The control routing topology partitioning reconstruction unit is connected to the global covariance matrix analysis unit. It is used to reconstruct the topology partitioning of the internal logical routing based on coupling affinity, thereby canceling physical interference between control loops. The logic rack bounce compensation unit is connected to the control route topology partition reconstruction unit. It is used to obtain the bounce amount based on the absolute value of the output difference before and after the drive mechanism switching during logic route reconstruction, and to send a limiting signal to the control command output boundary unit when the bounce amount exceeds the safety threshold. The control command output boundary unit is connected to the control routing topology partition reconstruction unit and the logic rack bounce compensation unit. It is used to limit the control gain change rate of the current partition to within -5% to 5% when a limiting signal is received, and output the decoupled control command flow.

[0006] Preferably, the global covariance matrix analysis unit includes an asynchronous phase alignment calibration operator and a discrete delay buffer chain; the asynchronous phase alignment calibration operator is used to extract the communication timestamps in the data streams uploaded by each control loop and calculate the transmission delay difference component between the communication timestamps and the local clock of the central processing unit; the discrete delay buffer chain is used to dynamically attach delay buffer data of corresponding length to the end of the data stream queue according to the transmission delay difference component, and to align the deviation data of each control loop incorporated into the current partition in phase through time domain occupancy.

[0007] Preferably, the global covariance matrix analysis unit includes a covariance matrix construction module and an eigenvector extraction module. The covariance matrix construction module is used to collect the feedback error change rate data stream of the distributed control loop in the current control cycle, calculate the covariance coefficient between each pair of control loops, fill the corresponding matrix elements of the global covariance matrix with the covariance coefficient, and establish a square matrix characterizing the nonlinear coupling degree of the multi-loop aerodynamic field, thermodynamic cycle, and hydraulic network. The eigenvector extraction module is connected to the covariance matrix construction module and is used to perform singular value decomposition on the global covariance matrix, solve for the principal component eigenvectors and associated eigenvalues, determine the interlocking partition boundaries between each pair of distributed control loops by comparing the weight distribution of the principal component eigenvectors, generate a topology weight mapping table to characterize the coupling association state of the multi-loop, and input the topology weight mapping table into the control route topology partition reconstruction unit.

[0008] Preferably, the control routing topology partitioning and reconstruction unit includes a topology decoupling partitioning module and a routing short-circuit reassembly module. The topology decoupling partitioning module receives the analysis results of the cross-covariance matrix of the global feedback error change rate, dynamically identifies the coupling affinity between distributed control loops in the information space, and classifies control loops with nonlinear mutual interference into the same interlocked control partition based on the coupling affinity. It independently allocates mutually decoupled control gain matrices within the interlocked control partition. The routing short-circuit reassembly module is connected to the topology decoupling partitioning module and is used to establish a dynamic cross-feedforward compensation path within the interlocked control partition. By rewriting the internal forwarding address of the controlled command stream, it drives the adaptive short-circuit reassembly of the logical routing within the control system, thus canceling the physical interaction interference between distributed control loops before the control command stream is issued.

[0009] Preferably, the control command output boundary unit includes an amplitude saturation clamping module and a rate of change dynamic limiting module; the amplitude saturation clamping module is used to monitor the absolute amplitude of the decoupled control command stream output by each control loop in real time, and clamp the control command stream within a safe boundary value when the absolute amplitude reaches the maximum output boundary; the rate of change dynamic limiting module is connected to the amplitude saturation clamping module and is used to receive the limiting activation signal input by the logic rack bounce compensation unit, and limit the control gain change rate of the current partition to a closed interval of -5% to 5%.

[0010] Preferably, the distributed control loop includes a frequency converter drive control loop, a heat pump source-side regulation loop, and a terminal valve opening control loop; the frequency converter drive control loop is connected to the frequency converter unit and is used to regulate the flow rate of hot and cold fluids within the building site; the heat pump source-side regulation loop is connected to the ground source heat pump regulating valve and is used to control the temperature difference of the heat exchange cycle of the ground source heat pump; the terminal valve opening control loop is connected to the electric regulating valve of the terminal air conditioning unit in the building site and is used to regulate the indoor air temperature; the data acquisition unit is connected to the frequency converter drive control loop, the heat pump source-side regulation loop, and the terminal valve opening control loop through a communication bus network to acquire the aerodynamic field deviation data, thermodynamic cycle deviation data, and hydraulic pipeline pressure fluctuation data generated by the mutual coupling of each control loop in the same physical field, and converts the deviation data and pressure fluctuation data into a feedback error change rate data stream.

[0011] Preferably, the data acquisition unit further includes an asynchronous packet loss compensator; the asynchronous packet loss compensator is used to extract the historical sequence of the feedback error change rate received in the previous sequence when data packet loss occurs in the communication bus network, and fill in the data lost in the current control cycle through a first-order forward differential extrapolation algorithm, so as to maintain the continuous evolution of the data flow in the global covariance matrix analysis unit.

[0012] Preferably, the logic rack bounce compensation unit further includes a drive mechanism aging tracking module; the drive mechanism aging tracking module is used to record the cumulative time series of logic rack bounce generated by the drive mechanism corresponding to each control loop at the moment of each topology reorganization during the system operation cycle, and calculate the time change rate of logic rack bounce with operating time as an aging decay index; when the aging decay index exceeds the predetermined safety threshold, the drive mechanism aging tracking module outputs a hardware warning signal and reduces the maximum allowable control gain of the corresponding control partition.

[0013] Preferably, the sampling control period for the data acquisition unit to acquire distributed control loop data is 10ms to 50ms; the maximum buffer time of the delay buffer chain set when the global cross-variance matrix analysis unit extracts timestamps for phase calibration is 5ms to 20ms; the action time window of the logic rack bounce compensation unit when limiting the rate of change of control gain is 100ms to 300ms, and the limit update step size of the rate of change of control gain is 1% to 3%.

[0014] A method for coordinated control of electromechanical energy conservation in building sites, used to operate a coordinated control system for electromechanical energy conservation in building sites, includes the following steps: Step S101: Obtain the feedback error change rate data stream and the corresponding communication timestamp for each control loop; Step S102: calibrate the phase of the feedback error change rate data stream according to the delay difference component of the communication timestamp, and construct the cross-covariance matrix of the feedback error change rate to identify the coupling affinity between control loops. Step S103: Reconstruct the topology partition of the internal logical routing based on coupling affinity to cancel out physical interference between control loops; Step S104: During logical route reorganization, the bounce amount is obtained based on the absolute value of the difference between the output before and after the switching of the drive mechanism, and a limiting signal is generated when the bounce amount exceeds the safety threshold. Step S105: Upon receiving the limiting signal, the control gain change rate of the current partition is limited to within -5% to 5%, and a decoupling control command stream is output.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. In the electromechanical energy-saving linkage control of building sites, the data acquisition unit extracts the deviation change rate vector, and uses the first derivative of the common mode isolation module to verify and isolate the topology misjudgment caused by external shocks; the system uses the correlation arbitration module to calculate the cross-covariance matrix of the deviation of each loop in the sliding window, and the partition reconstruction module compares the matrix with the threshold and removes the parallel topology, dividing the loops with coupling relationship into the same interlocked partition. At the same time, the internal bidirectional cross-compensation gain bus is activated and the deviation change rate is injected into the control gain formula as a feedforward term to realize the early offset of nonlinear interference between loops in the information space and block the cascade resonance of the regulation loop caused by the linkage of multiple source devices.

[0016] 2. When the control routing topology partition reconstruction unit reconstructs the control topology, the logic rack bounce compensation unit monitors the control output of the drive mechanism under the previous topology and the initial output under the new topology in real time. It obtains the bounce amount by the absolute value of the difference between the two and writes it as a dynamic integral attenuation factor into the integrator input of the current interlocked partition controller. When the bounce amount exceeds the safety mutation threshold, it forces the control gain change rate of the current partition to reduce the integral accumulation slope of the current controller, smoothly switch the transient gain evolution trajectory, and avoid the instruction jump caused by topology reorganization from inducing saturation zone control divergence and commutation wear of the drive mechanism.

[0017] 3. To address the technical deficiency of degraded cross-covariance matrix identification accuracy caused by time-domain interleaving of data streams in non-isochronous fieldbus networks, asynchronous phase alignment calibration operators configured at the tail end of each input channel of the data acquisition unit extract the communication timestamps in the data streams uploaded by each adjustment loop and calculate the transmission delay difference component between the timestamp and the local clock of the central processing unit. Based on the value of the delay difference component, a discrete delay buffer chain of corresponding length is dynamically attached to the tail end of the data stream queue. By occupying time-domain space, it is ensured that all deviation data incorporated into the interlocked partition are aligned in phase, thereby eliminating topology misjudgment caused by delay in non-ideal network environments and maintaining the steady state of continuous evolution of decoupled control gain flow. Attached Figure Description

[0018] Figure 1This is a flowchart of the multi-loop control routing topology reconfiguration process of the present invention; Figure 2 This is a system diagram of multi-loop control routing topology reconfiguration according to the present invention.

[0019] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0021] A building site electromechanical energy-saving linkage control system includes: The data acquisition unit is used to acquire the feedback error change rate data stream and the corresponding communication timestamp of each control loop; The global covariance matrix analysis unit, connected to the data acquisition unit, is used to calibrate the phase of the feedback error change rate data stream based on the delay difference component of the communication timestamp, and to construct the covariance matrix of the feedback error change rate to identify the coupling affinity between control loops. The control routing topology partitioning reconstruction unit is connected to the global covariance matrix analysis unit. It is used to reconstruct the topology partitioning of the internal logical routing based on coupling affinity, thereby canceling physical interference between control loops. The logic rack bounce compensation unit is connected to the control route topology partition reconstruction unit. It is used to obtain the bounce amount based on the absolute value of the output difference before and after the drive mechanism switching during logic route reconstruction, and to send a limiting signal to the control command output boundary unit when the bounce amount exceeds the safety threshold. The control command output boundary unit is connected to the control routing topology partition reconstruction unit and the logic rack bounce compensation unit. It is used to limit the control gain change rate of the current partition to within -5% to 5% when a limiting signal is received, and output the decoupled control command flow.

[0022] Preferably, the global covariance matrix analysis unit includes an asynchronous phase alignment calibration operator and a discrete delay buffer chain; the asynchronous phase alignment calibration operator is used to extract the communication timestamps in the data streams uploaded by each control loop and calculate the transmission delay difference component between the communication timestamps and the local clock of the central processing unit; the discrete delay buffer chain is used to dynamically attach delay buffer data of corresponding length to the end of the data stream queue according to the transmission delay difference component, and to align the deviation data of each control loop incorporated into the current partition in phase through time domain occupancy.

[0023] Preferably, the global covariance matrix analysis unit includes a covariance matrix construction module and an eigenvector extraction module. The covariance matrix construction module is used to collect the feedback error change rate data stream of the distributed control loop in the current control cycle, calculate the covariance coefficient between each pair of control loops, fill the corresponding matrix elements of the global covariance matrix with the covariance coefficient, and establish a square matrix characterizing the nonlinear coupling degree of the multi-loop aerodynamic field, thermodynamic cycle, and hydraulic network. The eigenvector extraction module is connected to the covariance matrix construction module and is used to perform singular value decomposition on the global covariance matrix, solve for the principal component eigenvectors and associated eigenvalues, determine the interlocking partition boundaries between each pair of distributed control loops by comparing the weight distribution of the principal component eigenvectors, generate a topology weight mapping table to characterize the coupling association state of the multi-loop, and input the topology weight mapping table into the control route topology partition reconstruction unit.

[0024] Preferably, the control routing topology partitioning and reconstruction unit includes a topology decoupling partitioning module and a routing short-circuit reassembly module. The topology decoupling partitioning module receives the analysis results of the cross-covariance matrix of the global feedback error change rate, dynamically identifies the coupling affinity between distributed control loops in the information space, and classifies control loops with nonlinear mutual interference into the same interlocked control partition based on the coupling affinity. It independently allocates mutually decoupled control gain matrices within the interlocked control partition. The routing short-circuit reassembly module is connected to the topology decoupling partitioning module and is used to establish a dynamic cross-feedforward compensation path within the interlocked control partition. By rewriting the internal forwarding address of the controlled command stream, it drives the adaptive short-circuit reassembly of the logical routing within the control system, thus canceling the physical interaction interference between distributed control loops before the control command stream is issued.

[0025] Preferably, the control command output boundary unit includes an amplitude saturation clamping module and a rate of change dynamic limiting module; the amplitude saturation clamping module is used to monitor the absolute amplitude of the decoupled control command stream output by each control loop in real time, and clamp the control command stream within a safe boundary value when the absolute amplitude reaches the maximum output boundary; the rate of change dynamic limiting module is connected to the amplitude saturation clamping module and is used to receive the limiting activation signal input by the logic rack bounce compensation unit, and limit the control gain change rate of the current partition to a closed interval of -5% to 5%.

[0026] Preferably, the distributed control loop includes a frequency converter drive control loop, a heat pump source-side regulation loop, and a terminal valve opening control loop; the frequency converter drive control loop is connected to the frequency converter unit and is used to regulate the flow rate of hot and cold fluids within the building site; the heat pump source-side regulation loop is connected to the ground source heat pump regulating valve and is used to control the temperature difference of the heat exchange cycle of the ground source heat pump; the terminal valve opening control loop is connected to the electric regulating valve of the terminal air conditioning unit in the building site and is used to regulate the indoor air temperature; the data acquisition unit is connected to the frequency converter drive control loop, the heat pump source-side regulation loop, and the terminal valve opening control loop through a communication bus network to acquire the aerodynamic field deviation data, thermodynamic cycle deviation data, and hydraulic pipeline pressure fluctuation data generated by the mutual coupling of each control loop in the same physical field, and converts the deviation data and pressure fluctuation data into a feedback error change rate data stream.

[0027] Preferably, the data acquisition unit further includes an asynchronous packet loss compensator; the asynchronous packet loss compensator is used to extract the historical sequence of the feedback error change rate received in the previous sequence when data packet loss occurs in the communication bus network, and fill in the data lost in the current control cycle through a first-order forward differential extrapolation algorithm, so as to maintain the continuous evolution of the data flow in the global covariance matrix analysis unit.

[0028] Preferably, the logic rack bounce compensation unit further includes a drive mechanism aging tracking module; the drive mechanism aging tracking module is used to record the cumulative time series of logic rack bounce generated by the drive mechanism corresponding to each control loop at the moment of each topology reorganization during the system operation cycle, and calculate the time change rate of logic rack bounce with operating time as an aging decay index; when the aging decay index exceeds the predetermined safety threshold, the drive mechanism aging tracking module outputs a hardware warning signal and reduces the maximum allowable control gain of the corresponding control partition.

[0029] Preferably, the sampling control period for the data acquisition unit to acquire distributed control loop data is 10ms to 50ms; the maximum buffer time of the delay buffer chain set when the global cross-variance matrix analysis unit extracts timestamps for phase calibration is 5ms to 20ms; the action time window of the logic rack bounce compensation unit when limiting the rate of change of control gain is 100ms to 300ms, and the limit update step size of the rate of change of control gain is 1% to 3%.

[0030] A method for coordinated control of electromechanical energy conservation in building sites, used to operate a coordinated control system for electromechanical energy conservation in building sites, includes the following steps: Step S101: Obtain the feedback error change rate data stream and the corresponding communication timestamp for each control loop; Step S102: calibrate the phase of the feedback error change rate data stream according to the delay difference component of the communication timestamp, and construct the cross-covariance matrix of the feedback error change rate to identify the coupling affinity between control loops. Step S103: Reconstruct the topology partition of the internal logical routing based on coupling affinity to cancel out physical interference between control loops; Step S104: During logical route reorganization, the bounce amount is obtained based on the absolute value of the difference between the output before and after the switching of the drive mechanism, and a limiting signal is generated when the bounce amount exceeds the safety threshold. Step S105: Upon receiving the limiting signal, the control gain change rate of the current partition is limited to within -5% to 5%, and a decoupling control command stream is output.

[0031] Example 1: In a building site electromechanical linkage control environment including variable frequency drive units, ground source heat pump regulating valves, and electric regulating valves of building site terminal air conditioning units, the coexistence of multiple drive mechanisms in the physical field leads to strong nonlinear coupling interference between the flow rates of hot and cold fluids, the temperature difference of heat exchange cycles, and the indoor air temperature control loops through aerodynamic fields, thermodynamic cycles, and hydraulic networks. When the variable frequency unit frequently reverses or the electric regulating valves are opened and closed transiently, the feedback physical parameters of adjacent loops strongly overlap in the time domain, causing control gain intermodulation. This physical coherence causes the control output of conventional independent regulating loops to be superimposed during linkage control, inducing continuous phase resonance. This leads to response lag and control energy capture in the dynamic field of hot and cold fluids and hydraulic pipelines, ultimately causing system-wide oscillations and mechanical fatigue and zero-point temperature drift damage to the underlying physical drive mechanism due to frequent reciprocating commutations. To address system instability caused by the aforementioned multi-loop coherence, the data acquisition unit acquires the feedback error rate of change data streams and corresponding communication timestamps of the frequency converter drive control loop, the heat pump source-side regulation loop, and the terminal valve opening control loop with a fixed sampling control cycle of 20ms. The asynchronous phase alignment calibration operator mounted inside the global covariance matrix analysis unit extracts the communication timestamps and calculates the transmission time between them and the local clock of the central processing unit. The delay difference component, a discrete delay buffer chain, dynamically attaches delay buffer data of corresponding length to the end of the data stream queue based on the value of the delay difference component during transmission. Time-domain occupancy aligns the deviation data of each control loop in the phase of the virtual time section, eliminating topology misjudgments caused by transmission delays in the non-isochronous bus network. Simultaneously, the common-mode interference isolation module intercepts and extracts the instantaneous first derivative of the global control deviation rate of change vector in real time. When the global first derivative synchronously exceeds the safety threshold, it is determined that there is an external sudden step-mode impact rather than intrinsic coupling between loops. The system then locks the current system topology and maintains the current logical topology state without switching. This isolates pseudo-coupling caused by external shocks at the source of the data flow. Furthermore, the common-mode interference frequency band isolation module calculates the difference between the global control deviation change rate vector of the current sampling period and the corresponding vector of the previous sampling period through the pre-discrete differential of the central processing unit, and divides it by the 20ms sampling period to obtain the instantaneous first derivative. When the absolute value of the first derivative is greater than the preset 50V safety threshold per second for three consecutive control periods, the module writes a set signal to the controller interrupt register, directly suspends the rewriting process of the control routing topology partition reconstruction unit, forcibly freezes all instruction forwarding logic in the routing table, and maintains the current logical topology state without switching.

[0032] Under standard operating conditions where the common-mode interference frequency band isolation module is not locked, the feedback error change rate data stream after phase alignment calibration flows into the cross-covariance matrix construction module of the global cross-covariance matrix analysis unit. The cross-covariance matrix construction module performs discrete product summation of the control deviations of each loop within a fixed-length sliding time window to calculate intermediate process indicators representing the transient coherence characteristics between loops; specifically based on the cross-covariance calculation formula... Construct the cross-covariance matrix of the feedback error rate of change, where, Let be the cross-covariance matrix of the feedback error rate of change. This is the global control deviation vector. This is the transpose matrix of the global control deviation vector. The total number of sampling points within the sliding time window is represented by this matrix, which serves as a square matrix characterizing the nonlinear coupling degree of the multi-loop aerodynamic field, thermodynamic cycle, and hydraulic network. Correspondingly, the feature vector extraction module performs singular value decomposition on the global cross-covariance matrix to calculate the principal component eigenvectors and associated eigenvalues. By comparing the weight distribution of the principal component eigenvectors, the interlocking partition boundaries between each pair of distributed control loops are uniquely determined. A topology weight mapping table characterizing the multi-loop coupling association state is output and sent to the control routing topology partitioning reconstruction unit. This is done to eliminate multi-physics field interference. Due to the significant difference between the response time constant and the measurement dimensions, before constructing the cross-covariance matrix, the system performs zero-mean normalization on the collected variable frequency flow rate, cyclic temperature difference rate, and pipeline pressure fluctuation rate. Specifically, it extracts the sample mean and standard deviation of each physical quantity data stream within the slip time window, subtracts the mean from the original value of each sampling point, and then divides by the standard deviation to convert it into a dimensionless standardized scalar. Based on the physical characteristic that the thermodynamic cycle lags behind the hydraulic pipeline flow rate change, time alignment of two control cycles is introduced in the time domain for the temperature difference data. Compensation is used to ensure that feature vectors with different physical properties have equal comparison weights under isomorphic sampling periods, eliminating high-frequency nonlinear measurement noise caused by fluctuations in hydraulic pipeline wall friction resistance and mechanical transmission clearance. Before calculating the elements of the cross-covariance matrix, the data acquisition unit calls the moving average filtering program to filter out high-frequency noise components. Based on the low-pass filtering theory of discrete time domain, the arithmetic mean of adjacent discrete data points in the sampling sequence is calculated to suppress random Gaussian noise. The specific steps include: selecting a 10-period sliding sampling window, the data acquisition unit captures the control loop data of the current control cycle, removes isolated abnormal data that exceed three times the standard deviation of the normal baseline, adds the 10 historical sampling data within the sliding time window and divides by 10 to obtain the smoothed control deviation change rate vector. The sampling window length N represents the total number of continuous discrete samples used for arithmetic mean calculation. The value determines the low-pass filter cutoff frequency and the system phase lag. If the window is too long, the control command response will be delayed. If it is too short, it will not be able to fully suppress high-frequency pressure pulsation noise interference. Smoothing processing pre-decouples multi-dimensional physical parameters at the input end.

[0033] The topology decoupling partitioning module within the control routing topology partitioning reconstruction unit receives the topology weight mapping table and scans the matrix elements in the cross-covariance matrix one by one in the virtual information space. When the absolute value of the matrix elements corresponding to any two independent control loops exceeds the preset critical coupling threshold of 0.15, a logic reconstruction judgment is triggered. The original independent parallel control flow topology of the corresponding regulation loop is decoupled, and the regulation loops with coupling relationships are forcibly assigned to the same interlocked control partition. Within the interlocked control partition, mutually decoupled control gain matrices are independently allocated. The routing short-circuit reassembly module establishes a dynamic cross-feedforward compensation path within the interlocked control partition. By rewriting the internal forwarding address of the controlled command flow, it drives the short-circuit reassembly of the logical routing within the control system. The instantaneous deviation rate of one loop is multiplied by a coefficient term as the feedforward dependent variable and directly added in-situ to the control gain calculation of the other loop. This allows the control disturbance of the preceding loop to be offset in-situ by the opposite control action of the subsequent loop within the information space before it is transformed into physical field coherent damage. This prevents the cascading resonance of the regulation loops caused by the linkage of multiple source devices. Specifically, the controlled command stream adopts the standard industrial Ethernet protocol message format, and its internal forwarding address is stored in the target node identification field and port image address field of the message header. The routing short-circuit reassembly module directly rewrites the device target logical number in this identification field, overwriting the target address originally sent to the independent control channel with the shared memory address of the cross-compensation nodes within the interlocked control partition. In the controller network protocol stack, the media access control layer directly resets the instruction stream, achieving adaptive short-circuit reconfiguration without requiring re-addressing and forwarding by the upper-layer application software. The central processing unit completes the instruction stream reset at the media access control layer and sends the reconfigured decoupled control instruction stream to the controlled hardware object register through the universal serial bus interface. Specifically, the decoupled control data stream temporarily stored in the shared memory address contains a 16-bit wide voltage duty cycle control word. The communication bus network distributes this control word to the corresponding zone frequency converter drive unit controller or electric regulating valve regulator. The regulator's internal digital-to-analog converter chip converts the 16-bit duty cycle control word into a 4mA to 20mA industrial standard current signal, changing the stepper motor drive inside the electric regulating valve. The voltage causes the stepper motor to produce an angular displacement opposite to the direction of the pressure fluctuation. The variable frequency drive unit controller converts the control word into a frequency adjustment command from 0 Hz to 50 Hz, changing the AC output frequency of the inverter bridge and adjusting the speed of the variable frequency pump. This allows the rewriting of the internal logical routing of the control network to be directly converted into a reverse neutralization action of the fluid kinetic energy within the physical pipeline network. At the instant the control routing topology partitioning reconstruction unit triggers the topology partitioning action, to eliminate the risk of discontinuous control command jumps caused by the transient discontinuity of the original state values ​​of the system controller's internal registers, the logic rack bounce compensation unit extracts in real time the last step output control quantity of the corresponding drive mechanism under the previous topology architecture and the control quantity initially calculated under the new topology partition, and calculates the bounce amount according to the bounce amount calculation formula. The absolute value of the difference between the two is determined to uniquely define the logic rack bounce, where, For logic rack bounce, This is the final output control variable in the preceding topology architecture. The initial calculation output is calculated under the new topology partition. When the logic rack bounce exceeds the safety mutation threshold, a limiting activation signal is sent to the control command output boundary unit. The dynamic limiting module of the control command output boundary unit forces the control gain change rate of the current partition to be limited to the closed interval of -5% to 5% within a time window of 100ms to 300ms, reducing the integral accumulation slope of the current controller to implement the dynamic yielding of the control oversaturation state. At the same time, the amplitude saturation clamping module monitors the absolute amplitude of the decoupled control command flow output of each control loop in real time, and clamps the control command flow within the safety boundary value when the absolute amplitude reaches the maximum output boundary, smoothly switching the transient gain evolution trajectory so that the reconstructed decoupled gain flow is converted into a standard control command that conforms to the underlying hardware communication protocol and distributed back to the physical drive mechanism. As the control command is transmitted back to the electromechanical linkage control environment of the building site, the reciprocating overshoot caused by physical coherence between loops converges within two to three adjustment cycles. The control route partition reorganization mechanism autonomously strips common-mode interference and cancels physical interference between loops in situ without the support of physical field sensors. The system employs a two-way constraint defense mechanism involving logic rack bounce and timestamp difference. This mechanism enables the system to forcibly block divergent changes in control commands when facing non-ideal operating conditions such as occasional packet loss in the field network or sudden changes in electromechanical load. The entire system is upgraded from a fragile state that relies on ideal networks and smooth operating conditions to one that possesses the ability to regulate transmission delays and transient saturation in situ. The entire building site electromechanical energy-saving linkage control system maintains a steady-state control effect with continuous evolution of the decoupled control gain flow in a multi-variable strongly coupled environment. In this process, address rewriting at the digital level is directly transformed into... The physical counter-impact action of the actuator, because the control system directly distributes the rewritten decoupled control command stream to the frequency converter and regulating valve of the adjacent control loop, causes the subsequent regulating valve to generate a valve displacement completely opposite to the direction of the preceding physical field fluctuation within 10ms of receiving the command. Through the active fine adjustment of the valve, a local pressure reverse counter-impact wave is generated at the fluid dynamics level, thereby directly neutralizing the mechanical energy impact caused by the transmission of aerodynamic field and hydraulic fluctuation within the physical pipeline network, completing the closed-loop control from virtual information routing to the energy dissipation of the underlying physical field.

[0034] Example 2: Under the verification condition of multi-controlled loop dynamic interference conditioning in the electromechanical energy-saving linkage control system of a building site, this experiment relies on a hardware-in-the-loop linkage simulation test platform composed of a variable frequency drive unit, a ground source heat pump regulating valve, and an electric regulating valve of the building site terminal air conditioning unit. The data source of the test platform is the actual physical operation monitoring system, which includes a data acquisition card with a sampling rate of 50Hz and a measurement resolution of 0.01V to generate a feedback error change rate data stream. When determining the value of the sampling control cycle, the constraints are the bandwidth limit of the distributed network bus and the variable frequency drive control loop. The transient response speed of the circuit and the setting of the sampling control cycle are used to balance the real-time performance of data acquisition and the computational load of the control center processor. By establishing a multiple matching rule between the sampling frequency and the highest physical interference frequency of the controlled loop, when the fluid coherence frequency of the monitored object increases and exhibits high-frequency oscillation characteristics in the time domain, in order to prevent bus data packet loss and control phase decoupling distortion, the sampling control cycle converges to the lower limit of its value range. Under rated load conditions, the sampling control cycle is specifically set to 20ms based on the upper limit of the fluid interference frequency envelope as a global benchmark input parameter for the multi-dimensional comparison system.

[0035] During the verification of the control flow, the experimental group using this technical solution operated under a 20ms sampling control cycle. Throughout the experiment, Gaussian white noise with a signal-to-noise ratio of 20dB and 50Hz power frequency interference harmonics generated by the power grid were introduced to simulate interference in the actual electromagnetic environment. Simultaneously, three different problem intensity gradient experimental groups were designed to correspond to low, medium, and high intensity hydraulic dynamic field coupling interference environments, respectively. In the high coupling interference environment, the unprocessed raw sensor-acquired signal generated interference glitch envelopes, and the feedback intermodulation distortion reached 28.4%. The asynchronous phase alignment calibration operator in the experimental group extracted data with communication timestamps and calculated... The transmission delay difference component was calculated to be 12.4ms. Based on this, the discrete delay buffer chain dynamically appended data of the corresponding length to the end of the data stream queue. Simultaneously, the common-mode interference isolation module calculated the instantaneous first derivative of the global control deviation change rate vector to be 35.6V / s. Since this value did not exceed the set impulse protection hard limit, the data stream flowed into the cross-covariance matrix construction module of the global cross-covariance matrix analysis unit. Within a sliding time window of 50 sampling points, the control deviations of each loop were discretely multiplied and superimposed to construct the cross-covariance matrix of the feedback error change rate. The intermediate process index representing the transient coherence characteristics between loops was calculated, specifically using the cross-covariance calculation formula. Produce the corresponding matrix elements, where, Let be the cross-covariance matrix of the feedback error rate of change. This is the global control deviation vector. This is the transpose matrix of the global control deviation vector. The total number of sampling points within the sliding time window is given. The weight of the principal component eigenvector obtained by the eigenvector extraction module is 0.246. A partially missing control group is set up in the multi-dimensional control system. Under the premise of maintaining the same hardware architecture and 20ms sampling control cycle, the dynamic cross feedforward compensation path and common mode interference frequency band isolation module are removed. Under the same 50Hz power frequency interference and 20dB Gaussian white noise impact, the partially missing control group cannot cancel the fluctuation transmission between control loops in advance in the information space. The reciprocating overshoot of its hot and cold fluid distribution flow rate is divergently superimposed in the time domain, which eventually produces a violent oscillation of the entire system. The experimental group, relying on the decoupling of the cross covariance matrix and the short-circuit reorganization action of the control routing topology partition reconstruction unit, reduces the feedback intermodulation distortion between loops from the original 28.4% to 2.1%. This provides factual evidence that the deep collaboration of each component module in the whole scheme produces a suppression of coherence beyond the sum of single features.

[0036] Two additional out-of-range control groups were set up in the multi-dimensional comparison system. The first out-of-range control group had a critical coupling threshold set to 0.05, which is lower than the lower limit of the limit range, and the second out-of-range control group had a critical coupling threshold set to 0.35, which is higher than the upper limit of the limit range. When the first out-of-range control group faced the aftershocks of pipeline pressure during operation, because the critical coupling threshold was lower than the working window, the topology decoupling partition module frequently triggered the logic reconstruction judgment. Within a 10-second time window, this caused the routing short-circuit reassembly module to change the instruction forwarding address 42 times. This topology switch caused the logical rack bounce calculated by the logical rack bounce compensation unit to increase the logical rack bounce amount. Exceeding the mutation limit, specifically according to the bounce calculation formula. Determine the value, where, For logic rack bounce, This is the final output control variable in the preceding topology architecture. The initial calculation output under the new topology partitioning caused the rate-of-change dynamic limiting module and amplitude saturation clamping module to fall into a continuous overmodulation state, resulting in excessive consumption of processor resources within the control system. When the second out-of-range control group faced multi-loop interference in the hydraulic pipeline network, due to the critical coupling threshold being set to 0.35, the topology decoupling partitioning module did not trigger the topology partitioning action within 120 seconds, causing the nonlinear interference between loops to superimpose in the physical field. The commutation reciprocating amplitude of the bottom variable frequency drive unit exceeded the standard by 41.5%, which in turn caused the drive actuator to experience temperature drift due to mechanical overload. In contrast, the test group using this technical solution and with the critical coupling threshold set to 0.15, under the same working conditions, its The topology reconfiguration action was triggered stably with only 3 partition reconfigurations. The control routing converged within 240ms, and the final output control variable change rate remained within a preset closed interval of -3.2% to 2.8%. The mechanical commutation wear of the underlying drive mechanism was reduced by 74.3%, and a critical coupling threshold of 0.15 was determined as the experimental basis for limiting the working window. Based on the relative gain matrix analysis principle in multivariable decoupling control, the nonlinear decoupling boundary was determined by calculating the energy gain distribution ratio brought about by the interaction of multiple loop physical fields. The specific process is as follows: Under zero-load operation, the data acquisition card continuously captures the static residual voltage of each control loop when there is no external disturbance, and calculates the root mean square value of the background noise to be 0.02 volts. The drive command is sent to the electric regulating valve in a step-by-step manner, and the pressure pulsation amplitude response sequence generated by the variable frequency flow control loop is recorded. The energy-related impedance moment of the fluid interference response is calculated. The maximum value of the energy-related impedance moment (10%) is added to the aforementioned root mean square value of the background noise to determine the critical coupling threshold as 0.15. The critical coupling threshold represents the minimum covariance correlation value that triggers the logical route reconstruction. The value range determines the activation sensitivity of the decoupling loop. When it is below 0.05, the inherent pulsation noise of the fluid in the physical pipeline network frequently crosses the judgment boundary, causing high-frequency, high-dissipation reorganization oscillations in the control route. When it is above 0.35, strong fluid coupling interference cannot be identified, resulting in cascade resonance overshoot, causing overheating and wear of the electric regulating valve's mechanical commutation. In this engineering design, the aforementioned upper limit of the control gain change rate of -5% to 5% and the critical coupling threshold of 0.15 have clear physical extreme boundary significance. When the absolute value of the control gain change rate is greater than 5%, the sudden change slope of the control voltage of the drive actuator in a single control cycle will directly cause the inverter bridge inside the underlying frequency converter to trigger transient overcurrent protection. When the gain change rate limit is too strict to an absolute value of less than 1%, the system will be unable to track the transient pressure fluctuations of the hydraulic pipeline due to excessive damping. Similarly, if the critical coupling threshold is set below 0.15, the inherent random Gaussian white noise of the fluid pipeline will frequently cross the decision line, causing the control route to generate high-frequency commutation oscillations and causing the processor to lock up. If the threshold is higher than 0.35. The strong coupling interference energy between the aerodynamic field and the thermodynamic loop cannot be identified in time, causing the overshoot of the cascaded resonance of multi-source equipment to directly lead to overheating of the actuator. Therefore, the range of values ​​for these two parameters is selected to maintain steady-state closed-loop balance. Based on the above multi-dimensional measured facts, the entire building site electromechanical energy-saving linkage control system provides a data closed loop for dynamic routing reorganization of the control topology and transient bounce limiting compensation in the linkage energy-saving control of multi-source electromechanical equipment. The system demonstrates the expected regulation effect in resisting physical interference, reducing actuator overshoot, and suppressing zero-point temperature drift.

[0037] Example 3: This example combines Figures 1 to 2 This document describes a building site electromechanical energy-saving linkage control system and method, such as... Figure 1 As shown, step S101 is executed to obtain the feedback error change rate data stream of each control loop and the corresponding communication timestamp. Step S102 is executed to calibrate the phase of the feedback error change rate data stream according to the delay difference component of the communication timestamp and construct the cross-covariance matrix of the feedback error change rate to identify the coupling affinity between control loops. Step S103 is executed to reconstruct the topology partition of the internal logic route according to the coupling affinity to cancel the physical interference between control loops. Step S104 is executed to obtain the bounce amount according to the absolute value of the output difference before and after the drive mechanism switching during the logic route reorganization, and generate a limiting signal when the bounce amount exceeds the safety threshold. Step S105 is executed to limit the control gain change rate of the current partition to within -5% to 5% when the limiting signal is received and output the decoupled control command stream.

[0038] like Figure 2 As shown, the central processing unit integrates a data acquisition unit, a global covariance matrix analysis unit, a control route topology partitioning reconstruction unit, a logic rack bounce compensation unit, and a control command output boundary unit. The data acquisition unit acquires the feedback error change rate data stream and timestamp; the global covariance matrix analysis unit calibrates the data stream phase and identifies loop coupling affinity; the control route topology partitioning reconstruction unit reconstructs the internal logical route and cancels physical interference between loops; the logic rack bounce compensation unit acquires the output differential bounce and sends a limiting protection signal; and the control command output boundary unit limits the control gain change rate and outputs a decoupled control command stream. The data acquisition unit receives input signals from a data acquisition card, which connects to a communication bus network for data aggregation. The processing logic within the central processing unit establishes communication interactions with the variable frequency unit, the ground source heat pump regulating valve, and the electric regulating valve via the communication bus network. The variable frequency unit performs the operation of regulating the flow rate of hot and cold fluids; the ground source heat pump regulating valve performs the operation of controlling the temperature difference of the heat exchange cycle; and the electric regulating valve performs the operation of regulating the indoor air temperature.

[0039] Example 4: When the system faces the overmodulation condition caused by the inertial time delay of the ground source heat pump distribution network and the dead zone of mechanical clearance friction of the control valve, the electric regulating valve of the air conditioning unit at the end of the building site will generate reciprocating overshoot during regulation. Due to the long-distance transport lag of the distribution fluid dynamic field, the feedback physical parameters of the terminal hot and cold fluid temperature control loop cannot reflect the real-time change of valve opening. As a result, the central processor of the control system continuously accumulates integral gain, which induces control oversaturation and the problem of floating initial value of integral saturation at the moment of topology route reorganization. Ultimately, the physical drive mechanism will cause continuous oscillation due to the accumulation of action delay and the control valve will be damaged by chatter.

[0040] To eliminate the aforementioned transmission and distribution delays and frictional losses, the system deploys adaptive parameter adjustment logic within the interlocked control partition established by the control routing topology partitioning reconstruction unit. Under initial state constraints, the data acquisition unit receives the original pressure deviation signal and valve displacement feedback code stream from the distributed control loop. The analog-to-digital conversion sampling bit width of the data acquisition card is set to 16 bits, and the transient input impedance is greater than 1MΩ to meet the microvolt-level electrical feature extraction accuracy requirements in non-ideal industrial high-noise environments. The parameter correction unit extracts the coupling affinity matrix elements corresponding to the current interlocked control partition from the global cross-covariance matrix within the process parameter calculation loop. According to the adaptive adjustment formula of control gain Calculate the gain correction amount, where, This is an adaptive correction term for controlling the gain. The preset scale conversion factor, To extract the corresponding cross-coupling variance components from the global covariance matrix, the routing short-circuit reassembly module receives the adaptive correction term and uses it as a multiplier operator to directly correct the dynamic feedforward gain within the interlocked control zone. Simultaneously, to address the second-order dissipation losses generated by long-distance hydraulic pipeline transport, the signal triggers a first-order discrete compensation loop. The quantization compensation loop uses a ring register to cache the control command residuals from the previous three control cycles and calculates the reverse hydraulic damping compensation gain based on the instantaneous negative polarity of the valve displacement sensor. This damping compensation gain is then directly accumulated in-situ and injected into the control command stream as a feedforward dependent variable. Before the control gain is issued, in-situ numerical offsetting of the inherent frictional dead zone and mass transfer inertial resistance of the terminal drive mechanism is completed. Specifically, the preset scale conversion factor is determined by dividing the maximum reference feedback voltage collected during the zero-load debugging phase of the system by the maximum limit eigenvalue of the cross-covariance matrix. Its specific value is fixed at 12V per unit variance. The calculation logic of the reverse hydraulic damping compensation gain is as follows: when the displacement sensor detects that the instantaneous speed of the valve is negative, it indicates that the valve is performing a closing deceleration action. The quantization compensation loop sequentially reads the arithmetic mean of the control command residuals of the past three control cycles cached in the ring register from the small end to the large end, multiplies it by a fixed damping image coefficient of 0.45 as the compensation gain term, and adds it to the current decoupled control command flow in place to completely offset the reverse impact force caused by the inertial lag of the fluid pipe flow.

[0041] As the control commands calculated by the parameter adaptive adjustment logic are sent to the electromechanical energy-saving linkage control system of the building site, the control saturation overmodulation frequency within the interlocked control zone attenuates. Without changing the hardware drive mechanism, the system decouples the control instability state intertwined with the mass transfer lag of the hydraulic pipeline and valve chatter through the parameter adaptive adjustment logic. By using the bidirectional physical constraint mechanism of bounce and instantaneous damping residual, control constraints are established at the register bottom layer. The tracking response error of the control valve is stably maintained within the rated allowable range of 1.5%. The entire system maintains the continuous evolution of the decoupled control gain flow and the reciprocating chatter wear depth of the bottom electric regulating valve is reduced.

[0042] Example 5: When the system faces the initial commissioning conditions of new building site deployment or batch changes of controlled equipment hardware, the central processing unit reads the static voltage code stream of the frequency converter unit under zero load state through the data acquisition card and calculates the variance as the background noise power value. At the same time, it increments the opening of the electric regulating valve and obtains the time-domain flow velocity disturbance sequence of the adjacent loop. It uses the cross-correlation algorithm to calculate the impedance correlation moment between the flow velocity disturbance and the valve action. 10% of the maximum value of the impedance correlation moment is added to the background noise power value. The resulting value is directly used as the critical coupling threshold of the interlock control zone in the current environment and written into the non-volatile memory as the boundary reference parameter of the real-time coherent decoupling control action.

[0043] When the linkage control system enters a long-term operating cycle and faces static baseline spatiotemporal drift caused by mechanical wear of the actuators and zero-point temperature drift, the timeliness guarantee update module initiates the calibration procedure. During low-load periods of the system, it sends step pulses to the actuators and monitors the transient decay time trend of pipeline hydraulic feedback and aerodynamic field fluctuations online. If the transient decay time deviates from the initial baseline by more than 5% for 48 consecutive control cycles, it is determined that the actuator has experienced characteristic degradation. The adaptive reconfiguration program automatically introduces a correction factor that decays logarithmically with time into the gain correction calculation, fine-tunes the critical judgment boundary of the topology decoupling partition module online to eliminate false interference components, and reconstructs the compensation accuracy of the routing control. This ensures that the entire building site electromechanical energy-saving linkage control system maintains a steady-state control effect with continuous evolution of control gain flow throughout its entire life cycle.

[0044] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0045] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A building site electromechanical energy-saving linkage control system, characterized in that, include: The data acquisition unit is used to acquire the feedback error change rate data stream and the corresponding communication timestamp of each control loop; The global covariance matrix analysis unit, connected to the data acquisition unit, is used to calibrate the phase of the feedback error change rate data stream based on the delay difference component of the communication timestamp, and to construct the covariance matrix of the feedback error change rate to identify the coupling affinity between control loops. The control routing topology partitioning reconstruction unit is connected to the global covariance matrix analysis unit. It is used to reconstruct the topology partitioning of the internal logical routing based on coupling affinity, thereby canceling physical interference between control loops. The logic rack bounce compensation unit is connected to the control route topology partition reconstruction unit. It is used to obtain the bounce amount based on the absolute value of the output difference before and after the drive mechanism switching during logic route reconstruction, and to send a limiting signal to the control command output boundary unit when the bounce amount exceeds the safety threshold. The control command output boundary unit is connected to the control routing topology partition reconstruction unit and the logic rack bounce compensation unit. It is used to limit the control gain change rate of the current partition to within -5% to 5% when a limiting signal is received, and output the decoupled control command flow.

2. The electromechanical energy-saving linkage control system for building sites according to claim 1, characterized in that, The global covariance matrix analysis unit includes an asynchronous phase alignment calibration operator and a discrete delay buffer chain; The asynchronous phase alignment calibration operator is used to extract the communication timestamps from the data streams uploaded by each control loop and to calculate the transmission delay difference component between the communication timestamps and the local clock of the central processing unit. Discrete delay buffer chains are used to dynamically attach delay buffer data of corresponding length to the end of the data stream queue according to the transmission delay difference components, and to align the deviation data of each control loop incorporated into the current partition in phase through time domain occupancy.

3. The electromechanical energy-saving linkage control system for building sites according to claim 1, characterized in that, The global covariance matrix analysis unit includes a covariance matrix construction module and an eigenvector extraction module. The covariance matrix construction module collects the feedback error change rate data stream of the distributed control loops in the current control cycle, calculates the covariance coefficients between each pair of control loops, fills the corresponding matrix elements of the global covariance matrix with the covariance coefficients, and establishes a square matrix characterizing the nonlinear coupling degree of the multi-loop aerodynamic field, thermodynamic cycle, and hydraulic network. The eigenvector extraction module is connected to the covariance matrix construction module and is used to perform singular value decomposition on the global covariance matrix, calculate the principal component eigenvectors and associated eigenvalues, determine the interlocking partition boundaries between each pair of distributed control loops by comparing the weight distribution of the principal component eigenvectors, generate a topology weight mapping table to characterize the coupling association state of the multi-loops, and input the topology weight mapping table into the control route topology partition reconstruction unit.

4. The electromechanical energy-saving linkage control system for building sites according to claim 1, characterized in that, The control routing topology partitioning and reconstructing unit includes a topology decoupling partitioning module and a routing short-circuit reconstructing module; The topology decoupling partitioning module receives the analysis results of the cross-covariance matrix of the global feedback error change rate, dynamically identifies the coupling affinity between distributed control loops in the information space, and classifies control loops with nonlinear mutual interference into the same interlocked control partition based on the coupling affinity. Within the interlocked control partition, it independently allocates the mutually decoupled control gain matrix. The route short-circuit reassembly module is connected to the topology decoupling partitioning module and is used to establish a dynamic cross-feedforward compensation path within the interlocked control partition. By rewriting the internal forwarding address of the controlled command stream, it drives the adaptive short-circuit reassembly of the logical routing within the control system, thus canceling the physical interaction interference between distributed control loops before the control command stream is issued.

5. The electromechanical energy-saving linkage control system for building sites according to claim 1, characterized in that, The control command output boundary unit includes an amplitude saturation clamping module and a rate of change dynamic limiting module. The amplitude saturation clamping module is used to monitor the absolute amplitude of the decoupled control command stream output by each control loop in real time, and clamp the control command stream within the safe boundary value when the absolute amplitude reaches the maximum output boundary. The rate of change dynamic limiting module is connected to the amplitude saturation clamping module and is used to receive the limiting activation signal input by the logic rack bounce compensation unit, and limit the control gain change rate of the current partition to a closed interval of -5% to 5%.

6. The electromechanical energy-saving linkage control system for building sites according to claim 1, characterized in that, The distributed control loop includes a variable frequency drive control loop, a heat pump source-side regulation loop, and a terminal valve opening control loop. The variable frequency drive control loop is connected to the variable frequency unit and is used to regulate the flow rate of hot and cold fluids within the building site. The heat pump source-side regulation loop is connected to the ground source heat pump regulating valve and is used to control the temperature difference of the heat exchange cycle of the ground source heat pump. The terminal valve opening control loop is connected to the electric regulating valve of the terminal air conditioning unit in the building site and is used to regulate the indoor air temperature. The data acquisition unit is connected to the variable frequency drive control loop, the heat pump source-side regulation loop, and the terminal valve opening control loop through a communication bus network to acquire aerodynamic field deviation data, thermodynamic cycle deviation data, and hydraulic pipeline pressure fluctuation data generated by the mutual coupling of each control loop in the same physical field, and converts the deviation data and pressure fluctuation data into a feedback error change rate data stream.

7. The electromechanical energy-saving linkage control system for building sites according to claim 1, characterized in that, The data acquisition unit also includes an asynchronous packet loss compensator; the asynchronous packet loss compensator is used to extract the historical sequence of the feedback error change rate received in the previous sequence when data packet loss occurs in the communication bus network, and fill in the data lost in the current control cycle through the first-order forward differential extrapolation algorithm, so as to maintain the continuous evolution of the data flow in the global covariance matrix analysis unit.

8. The electromechanical energy-saving linkage control system for building sites according to claim 1, characterized in that, The logic rack bounce compensation unit also includes a drive mechanism aging tracking module. The drive mechanism aging tracking module is used to record the cumulative time series of logic rack bounce generated by the drive mechanism corresponding to each control loop at the moment of each topology reorganization during the system operation cycle, and calculate the time change rate of logic rack bounce with operating time as an aging decay index. When the aging decay index exceeds the predetermined safety threshold, the drive mechanism aging tracking module outputs a hardware warning signal and reduces the maximum allowable control gain of the corresponding control partition.

9. The electromechanical energy-saving linkage control system for building sites according to claim 1, characterized in that, The sampling control cycle for the data acquisition unit to acquire distributed control loop data is 10ms to 50ms; the maximum buffer time of the delay buffer chain set by the global covariance matrix analysis unit when extracting timestamps for phase calibration is 5ms to 20ms; the action time window of the logic rack bounce compensation unit when limiting the rate of change of control gain is 100ms to 300ms, and the limit update step of the rate of change of control gain is 1% to 3%.

10. A method for electromechanical energy-saving linkage control of a building site, used to operate the electromechanical energy-saving linkage control system of claim 1, characterized in that, Includes the following steps: Step S101: Obtain the feedback error change rate data stream and the corresponding communication timestamp for each control loop; Step S102: calibrate the phase of the feedback error change rate data stream according to the delay difference component of the communication timestamp, and construct the cross-covariance matrix of the feedback error change rate to identify the coupling affinity between control loops. Step S103: Reconstruct the topology partition of the internal logical routing based on coupling affinity to cancel out physical interference between control loops; Step S104: During logical route reorganization, the bounce amount is obtained based on the absolute value of the difference between the output before and after the switching of the drive mechanism, and a limiting signal is generated when the bounce amount exceeds the safety threshold. Step S105: Upon receiving the limiting signal, the control gain change rate of the current partition is limited to within -5% to 5%, and a decoupling control command stream is output.

Images